Portal V6.1 Performance Tuning Open Mic Call on September 23, 2009 - Websphere

This is a discussion on Portal V6.1 Performance Tuning Open Mic Call on September 23, 2009 - Websphere ; Doug14, Logging can be very, or little important to the portal performance. This mainly depends on the amount of logging that you are using. A test to see how much this influences your portal is to remove all the logging ...

+ Reply to Thread
Page 2 of 2 FirstFirst 1 2
Results 21 to 36 of 36

Thread: Portal V6.1 Performance Tuning Open Mic Call on September 23, 2009

  1. Re: Portal V6.1 Performance Tuning Open Mic Call on September 23,2009

    Doug14,

    Logging can be very, or little important to the portal performance. This mainly depends on the amount of logging that you are using. A test to see how much this influences your portal is to remove all the logging and test accessing the same page that you were using before removing the logging. To get a true test, restart the server, and then hit the page once to pre-populate all the caches in portal. Open a new browser, then time the loading of that page. Then remove all the logging and reproduce.

    Use this URL for your logging references:
    http://www-01.ibm.com/support/docvie...id=swg21399563

    - Matt Munse

    The postings on this site are my own and do not necessarily represent the positions, strategies or opinions of IBM.

  2. Re: Portal V6.1 Performance Tuning Open Mic Call on September 23,2009

    Doug14,

    This is not true, caching is still useful.

    - Matt Munse

    The postings on this site are my own and do not necessarily represent the positions, strategies or opinions of IBM.

  3. Re: Portal V6.1 Performance Tuning Open Mic Call on September 23,2009

    Doug14,

    The performance of removing the conditions in the theme are dependent on if you're using CSA or SSA, however, the theme policies should be loaded, and the conditions should be very minimal in this condition.

    - Matt Munse

    The postings on this site are my own and do not necessarily represent the positions, strategies or opinions of IBM.

  4. Re: Portal V6.1 Performance Tuning Open Mic Call on September 23,2009

    Thanks for your very helpful reply.

    Regarding threads, should the number of threads for the
    WebContainer = threads for WorkManager.wpsWorkManager?

    That is why I'm asking about threading on requests. If multiple threads are spawned by every request, are there other areas to adjust?

  5. Re: Portal V6.1 Performance Tuning Open Mic Call on September 23,2009

    Proper database maintenance is integral to overall portal performance. See the tuning guide for specific settings or command synthax
    [IBM WebSphere Portal Version 6.1 Tuning Guide|http://www-01.ibm.com/support/docvie...d=swg27013972]

    Each database (DB2,ORACLE,ETC) may specific maintenance operations and the DBA should be able to help with that
    e.g. on DB2 you may want to frequently update database catalog statistics using runtstats ... once a week
    Also database reorg can maintain good physical organization of the data in the tables, ie defrag tables , etc to ensure efficient database access

    Index addition various database indexes are provided out ofteh box with portal . however depending on your environment and usage, you may need to create ad

  6. Re: Portal V6.1 Performance Tuning Open Mic Call on September 23,2009

    Proper database maintenance is integral to overall portal performance. See the tuning guide for specific settings or command synthax
    [IBM WebSphere Portal Version 6.1 Tuning Guide|http://www-01.ibm.com/support/docvie...d=swg27013972]

    Each database (DB2,ORACLE,ETC) may specific maintenance operations and the DBA should be able to help with that
    e.g. on DB2 you may want to frequently update database catalog statistics using runtstats ... once a week
    Also database reorg can maintain good physical organization of the data in the tables, ie defrag tables , etc to ensure efficient database access

    Index addition various database indexes are provided out ofteh box with portal . however depending on your environment and usage, you may need to create ad

  7. Re: Portal V6.1 Performance Tuning Open Mic Call on September 23,2009

    VivekAgarwal2005,

    For WCM, there's delete, and there's also purge. The deleted items can be un-deleted, this is why they're not removed.

    The API for a purge is in the jdocs for 6.1, and you would want to try on a test system to see if they work for 6.0.

    - Matt Munse

    The postings on this site are my own and do not necessarily represent the positions, strategies or opinions of IBM.

  8. Re: Portal V6.1 Performance Tuning Open Mic Call on September 23,2009

    VivekAgarwal2005,

    In 6.1 WCM uses a JCR DB and doesn't have the limitations on the number of content that were seen in 5.1. You can improve perf. by seperating your content into different site areas within a site if it makes sense for you, thus, less of a schema to traverse.

    - Matt Munse

    The postings on this site are my own and do not necessarily represent the positions, strategies or opinions of IBM.

  9. Re: Portal V6.1 Performance Tuning Open Mic Call on September 23,2009

    jocariz,

    The rest of the memory is due to native threads, and the libraries that the JVM is using. You are probably using a large number of database calls, and they are taking up the native memory. If you are passing large SQL queries and a lot of them this could be the case. Tuning to the portal itself should be able to help with this but a PMR to review your system settings will need to be opened.

    - Matt Munse

    The postings on this site are my own and do not necessarily represent the positions, strategies or opinions of IBM.

  10. Re: Portal V6.1 Performance Tuning Open Mic Call on September 23,2009

    PeterK2,

    There are three threads per request, one for the portal, one for the theme, and one to the portlet. This is true for 6.0, and for 6.1 this may differ for CSA.

    Also, yes, this is related to the number of threads in the pool, and why this should be set.

    - Matt Munse

    The postings on this site are my own and do not necessarily represent the positions, strategies or opinions of IBM.

  11. Re: Portal V6.1 Performance Tuning Open Mic Call on September 23,2009

    Hi,

    Can anyone take up this question ??

    Thanks

  12. Re: Portal V6.1 Performance Tuning Open Mic Call on September 23,2009

    (1) I learnt from the RedBook ‘Mastering DynaCache in WebSphere Commerce’ that DRS (Data Replication Service) is used for DynaCache Replication in a clustered environment. The book says:
    - In WAS version 5.x, the underlying transport used is Java message Service (JMS) messages, whereas
    - In WAS 6.x, the underlying mechanism has been rewritten using a Proprietary Transport to move data. This reduces overhead and improves performance.

    Can you help me understand the proprietary mechanism? Does NodeAgent have any role to play in Data Replication? I mean, is it necessary for a nodeagent to be running for DRS to function (in turn, Cache Replication to happen)?

    (2) Chapter7 of WP_Tuning_guide says: we can set some JVM CustomProperties to reduce the number & size of dynacache messages sent between nodes.
    Do we have to set the same properties on all the servers in the cluster? I mean, what is their scope? (i.e. Cluster scoped/Server scoped). Please explain in detail.

    (3) Just like you can supply cacheinstance.properties when using DynaCache programmatically, Is there any way to specify which cache-instance will be used for a in cachespec.xml?

  13. Re: Portal V6.1 Performance Tuning Open Mic Call on September 23,2009

    (1) I learnt from the RedBook ‘Mastering DynaCache in WebSphere Commerce’ that DRS (Data Replication Service) is used for DynaCache Replication in a clustered environment. The book says:
    - In WAS version 5.x, the underlying transport used is Java message Service (JMS) messages, whereas
    - In WAS 6.x, the underlying mechanism has been rewritten using a Proprietary Transport to move data. This reduces overhead and improves performance.

    Can you help me understand the proprietary mechanism? Does NodeAgent have any role to play in Data Replication? I mean, is it necessary for a nodeagent to be running for DRS to function (in turn, Cache Replication to happen)?

    (2) Chapter7 of WP_Tuning_guide says: we can set some JVM CustomProperties to reduce the number & size of dynacache messages sent between nodes.
    Do we have to set the same properties on all the servers in the cluster? I mean, what is their scope? (i.e. Cluster scoped/Server scoped). Please explain in detail.

    (3) Just like you can supply cacheinstance.properties when using DynaCache programmatically, Is there any way to specify which cache-instance will be used for a in cachespec.xml?

  14. Re: Portal V6.1 Performance Tuning Open Mic Call on September 23,2009

    Sorry for re-posting. Can't find a way to delete the duplicate.
    Admin,

  15. Re: Portal V6.1 Performance Tuning Open Mic Call on September 23,2009

    Can you tell us how to tune the thread pools?

    Should the WebContainer thread pool = WorkManager.wpsWorkManager thread pool?

    What does the WorkManager.wpsWorkManager do?

  16. Re: Portal V6.1 Performance Tuning Open Mic Call on September 23,2009

    Did you miss the recent open mic on WebSphere Portal V6.1 Performance Tuning? Please download the replay and listen at your leisure, [http://www.ibm.com/support/docview.wss?uid=swg21398723]

+ Reply to Thread
Page 2 of 2 FirstFirst 1 2